There were those fashionistas who would pop into shops to spot potential bargains before the sales. There were those who spotted items online, popped into the shop to touch the fabric, check the stitching or make sure they liked the item and ordered it online. Soon there will be those who will no longer leave their Gemini or ChatGPT to access huge "virtual Galeries Lafayette" based on a few words. "Buy me this pair of black shoes in my size that I like for less than 70 euros", they will ask a specific business agent, no less glamorous than a personal shopper. And the application will suggest a model, a price and a delivery date, allowing a discussion about the product without having to go back and forth between the brand's shops, whether physical or virtual. The end of the pleasure of shopping? Perhaps, but a growing demand from Internet users, according to some forty studies.

According to the Capgemini Research Institute, 64% of consumers say they feel overwhelmed by the complexity of digital purchasing paths and 58% say they are ready to delegate some of their decisions to automated tools. For retailers, the stakes are as much commercial as operational: Deloitte estimates that more than 70% of major retailers are now testing or deploying AI bricks capable of intervening in the customer journey or operations management.

In beauty, Sephora is an advanced laboratory. For several years now, the brand has been using AI and augmented reality tools to enable virtual product trials, skin diagnostics and shade matching. According to several industry studies reported by BrandXR and Netguru, the use of virtual try-ons can generate conversion increases of between 20% and over 30% in certain pilot shops, and up to 90% higher conversion rates among customers who use these tools than those who do not. For Sephora, the gain is twofold: a faster purchasing decision and a reduction in uncertainty about products where mistakes cost a lot in returns and dissatisfaction.

In fashion, Zara illustrates another approach to AI, less visible but just as structuring. The Inditex retailer uses algorithms to analyse sales, customer behaviour and demand signals in order to optimise stocks and assortments. This data-driven approach helps to reduce overstocking and improve collection rotation, a key issue in a sector where margins are sensitive. According to analyses by Deloitte and the NRF, predictive stock optimisation using AI can reduce the costs associated with unsold stock by 10-20%. Zara is also testing virtual fitting and look visualisation features, with the aim of improving the online experience and limiting returns, which represent a growing cost for textile e-commerce.

At Ikea, artificial intelligence and augmented reality are being used to remove a major barrier to purchase: the difficulty in projecting oneself. The Ikea Place application allows users to view full-scale furniture in their own homes. According to academic and sector analyses, augmented reality visualisation significantly improves confidence in the purchasing decision and helps to reduce returns, a particularly high cost item in the furniture sector. Deloitte estimates that returns can represent up to 20% of the value of sales in certain e-commerce segments, which explains the economic interest of these tools.

50 billion product data sheets online and up to date

It is in this context that Google's initiative takes on its full meaning. By proposing an open protocol enabling AI agents to interact with retailers' systems - catalogues, stocks, payments - Google is seeking to transform these still fragmented uses into a seamless experience. The aim is to reduce shopping basket abandonment, which still averages almost 70% according to industry benchmark studies, by bringing the intention formulated in a conversational interface closer to the actual purchase. During his presentation, Google CEO Sundar Pichai said, "Google search has always been an essential starting point for online shopping. Since 2021, our Shopping Graph has been providing information on products in real time. It lists more than 50 billion products, with their stocks, prices and reviews. More than 2 billion of these listings are updated every hour. Information at the speed of commerce"

For brands, the implications are strategic. Being present in these new pathways means structuring reliable product data that can be used by AI agents. According to Finch and Netguru, retailers whose catalogues are not optimised for generative engines risk losing up to 30% of their visibility in automated recommendations. Conversely, those that have mastered these tools can expect measurable gains: Deloitte and Retail Customer Experience observe average basket value increases of 15-25% when AI is integrated across the entire customer journey.

For now, customers will be able to pay from Google and Paypal without difficulty, but given the financial partners, such as Mastercard, Visa, Stripe or Adyen, the payment experience will become even more widespread.

This is no innocent step for Google, which has made its fortune on search and the ability of the American giant to sell advertising to brands and companies when faced with a web user looking for information. With direct purchase in Gemini, there is no doubt that brands will have the same opportunities to promote themselves.